By Minji Kim, Alice Lu, and Rudy Tanzi.
The third day of the meeting started with a session on presenilin and γ-secretase. Dennis Selkoe, Harvard Medical School, addressed the three principal stages of the life of Aβ: production, degradation, and aggregation. With respect to Aβ production, Selkoe reported that he and his colleagues were able to purify a γ-secretase complex from mammalian cells that produced physiological ratios of Aβ40 and Aβ42. Using electron microscopy and single particle image analysis, they performed structural analysis of the presenilin/γ-secretase complex. The 3-D reconstruction revealed that γ-secretase is a 100-angstrom particle with a central chamber of irregular diameter and two openings, one up top and one on the side. The structure looked like an oblong bead resembling a mini-proteasome, possibly made up of two presenilin (PS) molecules.
Regarding Aβ degradation, Selkoe showed that amyloid-β precursor protein (APP) transgenic mice lacking neprilysin (NEP) had increased amyloid plaque burden and Aβ immunoreactivity. The NEP-negative mice also developed amyloid angiopathy and had increased levels of Aβ40 and Aβ42. Selkoe also presented evidence for Aβ oligomers in flotillin-rich lipid rafts. He demonstrated that soluble Aβ oligomers, but not monomers, abrogated hippocampal LTP, in vivo and in vitro, and that this could be prevented by both active and passive immunotherapy and by small, orally bioavailable compounds (see Townsend below). Soluble Aβ oligomers could also transiently interfere with the memory of learned tasks in animals, and they caused loss of dendritic spines in organotypic hippocampal slice cultures. The spines could be rescued after 10 days with an Aβ-specific antibody (6E10).
Finally, Selkoe presented results of in utero electroporation of APP RNAi into rat cortex at E13. Knockdown of APP at this stage of development blocked neuronal migration to the cortical plate and led to retention in the intermediate zone. Similar results were obtained for knockdown of APP-like proteins APLP1 and APLP2. Interestingly, migration could be restored with APP C-terminal constructs, suggesting a role for APP adapters, for example, Fe65 family proteins.
Christian Haass, University of Munich, Germany, employed a γ-secretase reporter assay to show that when APP was retained in the endoplasmic reticulum (ER) or trans-Golgi network (TGN)—using brefeldin A or monensin, respectively), there was no cleavage by γ-secretase. Similar results were obtained for tannic acid, which prevents fusion of synaptic vesicles with the plasma membrane. Inhibition of endocytosis with a dominant negative mutant of dynamin was used to show that γ-secretase cleavage mainly takes place at the plasma membrane and in endosomes.
Next, Haass presented the identification of a conserved protease motif, GxGD, in various aspartyl proteases including the nematode presenilin homologue Spe4. A PS1/Spe4 active site chimera cleaved APP, but not Notch. He showed that a single amino acid at position x of the GxGD active site motif of PS is implicated in APP/Notch substrate selection of γ-secretase, and that final substrate selection of γ-secretase may occur very close to the active site. If the position between the two glycines is filled by phenylalanine, Notch intracellular domain (NICD) generation is blocked while AICD production from APP is allowed. If leucine lies between the glycines, both Notch and APP are cleaved. He pointed out that all signal peptide peptidases have GxGD motifs. Mutagenesis of the aspartate residue within the GxGD motif of signal peptide peptidases SPPL2 and SPPL3 in zebra fish resulted in a loss-of-function phenotype, indicating that SPPLs are aspartyl proteases of the GxGD family. Finally, he identified the type 2 transmembrane protein, TNFα, as a substrate for SPPL2. SPPL2 performs a γ-secretase-like dual intramembrane domain cleavage, which produces both an ICD-like fragment and a secreted C-terminal fragment.
Bart De Strooper, Center for Human Genetics, KU Leuven, Belgium, presented a nine transmembrane model for presenilin, and demonstrated that at least four different γ-secretase complexes exist. The complexes have different biological functions: Aph1A-/- mice had impaired embryological development, while Aph1BC-/- showed a normal phenotype—Aph1 is one of the subunits of γ-secretase. Interestingly, Aph1BC-/- mice had impaired γ-secretase activity in the adult brain, as well as impaired sensory gating (startle response), suggesting that the Aph1BC complex is a potentially interesting drug target for AD and that Aph1BC may have an important role in the development of dopaminergic pathways in the CNS.
The γ-secretase complexes also have different biochemical properties: Other γ-secretase subunits were affected in Aph1ABC-/- mice—there was no maturation of nicastrin (another γ-secretase component), no endoproteolysis of PS, and decreased levels of the fourth γ-secretase member, Pen-2. De Strooper also warned that a commercially available FRET-based assay for γ-secretase activity is not selective for γ-secretase. In studies of the role of the AICD, De Strooper stated that he could not confirm the recently reported effects of AICD on neprilysin expression whether using PS double knockouts, γ-secretase-inhibitors, or APP/APLP knockout mice (see ARF related news story).
Jie Shen, Brigham and Women’s Hospital and Harvard Medical School, Boston, described the use of Cre/loxP technology to generate presenilin double conditional double knockout mice, FB-PS cDKO, in which inactivation of presenilins was restricted spatially and temporally to the adult cerebral cortex. These FB-PS cDKO mice had mild memory deficits at 2 months with normal performance in rotarod and open field tests. However, at 6 months, they had severely impaired spatial learning and memory and severely impaired hippocampus- and amygdala-dependent memory, indicating that with age, memory impairment worsens. At 2 months, FB-PS cDKO also had short-term plasticity deficits: impaired LTP, impaired paired LTP, and reduced NMDA responses. Shen also showed an age-dependent impairment of NMDA receptor (NMDAR) responses in FB-PS cDKO mice. She proposed that presynaptic PS1 is a trans-synaptic regulator of postsynaptic NMDAR. Finally, She stated her preference for the hypothesis that FAD mutations in PS cause AD by attenuating PS function at least partially in an Aβ-independent manner (see ARF related news story).
Homira Behbahani, Karolinska Institute, Stockholm, discussed the role of PS1 and PS2 on mitochondrial membrane potential and oxygen consumption in mice embryo. She showed that in mouse embryonic fibroblast (MEF), there was a higher fractional area of mitochondria in PS1-/- cells, reduced COX1 levels in PS2-/- and DKO, low mitochondrial membrane potential in PS2-/- and DKO cells, and a low basal respiratory rate in PS2-/- cells.
Taisuke Tomita, University of Tokyo, presented the identification of PS1 as a molecular target of potent dipeptidic inhibitors, DAPT, CE, and DBZ. He demonstrated that DAPT specifically bound PS1 C-terminal fragment (CTF), but not the N-terminal fragment (NTF), that CE and DBZ bound to PS1 NTF, but not CTF, that CE and DBZ also bound to SPP and other polypeptides. He concluded that differences in the mode of binding might reflect enzyme specificity of the inhibitor (see ARF related news story).